Data Mining Can Cut Research Times, Experts Say
"If the financial markets operated like the healthcare market does with data, you would have to call 100 people every day to find out how the market is doing," said Greg Simon, CEO of FasterCures. The need for the pharmaceutical industry to make better use of large existing databases was the theme of a panel discussion Simon moderated at the World Health Care Congress in Washington, D.C. on April 20.
The need for better use of data is being driven by the move from blockbuster drugs to products targeted at narrower patient populations, increased emphasis on safety, cost effectiveness and patient outcomes, noted Glenn Bilawsky, CEO of i3. He said the pharma industry currently gets its data from three general categories: clinical trials; physician surveys; and retrospective claims and electronic medical records. However, all have their limitations and under the current system it is difficult or impossible to aggregate them, he said.
"There are two basic needs: the speed and timeliness of information, and for real-world data to be converted to usable information," Bilawsky added.
Tapping Insurers' Databases
Novartis Pharmaceuticals is extracting data from insurance claim databases and electronic medical records for a number of purposes, said Joseph DiCesare, the company's head of health economics and outcomes research. The company is mining the data for information on:
The total monetary burden imposed by diseases; Drug and other healthcare utilization patterns; Compliance and persistence; Major clinical outcomes, such as death or the need for surgery.
The databases Novartis is making use of are characterized by their large sample sizes, inclusion of real-world data and lengthy follow-up studies. DiCesare noted that extracting the right kind of information from these databases can help shorten research times when it comes to developing new drugs. At the same time, he acknowledged limitations on the clinical data, such as the lack of continuity of information on patients who tend to join and drop out of health insurance plans on a frequent basis, and "plant-specific biases."
Assessing the total value of a drug means studying a patient's total healthcare usage, noted William Crown, president, i3 Magnifi. As an example, he cited the introduction of selective serotonin reuptake inhibitors (SSRIs) -- antidepressants such as Prozac and Zoloft. These were about 10 times more expensive than the earlier tricyclic antidepressants, he told PIR. "But in total costs, we found [insurers] saved $2,000 per year per patient with SSRIs."
"The big difference was in the nonmental health costs," he continued. "Patients who started on SSRIs had better persistence [stayed on the drug longer], partly because the side effects were not as severe. The underlying depression was treated, and it relieved their other symptoms, like lower back pain and sleep difficulties."
The database belonging to UnitedHealth Care, the country's largest health insurer, is the starting point for i3 Magnifi to run data-mining studies like this, Crown said. "We are constantly looking for ways to make the database larger and richer, so we expand from UnitedHealth and include laboratory data." Smaller pharma companies don't have the resources to do large patient outcome studies like this themselves, Crown noted, yet these can be crucial for such things as discovering alternate markets for drugs.
The FDA is also aware of the importance of leveraging healthcare data, having launched a major effort to improve its public communication about drug risks following an announcement by HHS Secretary Mike Leavitt in February 2005. This has included the issuing of information sheets and online data for patients and healthcare professionals, said Susan Cummins, executive director of the Drug Safety Oversight Board at the FDA's Center for Drug Evaluation and Research. "The goal is to make the agency as transparent as we can be," she said.
"This is not just clinical trial data," Cummins added. In some circumstances, she said, the agency is reporting on adverse events in late Phase II clinical trials, or in the early post-approval phase just after the drug has been released onto the market. "We always evaluate new data in the context of everything else we know about a product," she said. -- Martin Gidron (mailto:mgidron@fdanews.com)